Department of Radiology, Beth Israel Deaconess Medical Center, Boston, MA, 02215, USA.
Harvard Medical School, Boston, MA, 02215, USA.
Med Phys. 2020 Jun;47(5):2139-2149. doi: 10.1002/mp.14102. Epub 2020 Mar 21.
Water equivalent diameter (WED) is a robust patient-size descriptor. Localizer-based WED estimation is less sensitive to truncation errors resulting from limited field of view, and produces WED estimates at different locations within one localizer radiograph, prior to the initiation of axial scans. This method is considered difficult to implement by the clinical community due to the necessary calibration between localizer pixel values (LPV) and attenuation, and the unknown stability of calibration results across scanners and over time. We investigated the stability of calibration results across 25 computed tomography (CT) scanners from three medical centers, and their stability over 3 ∼ 29 months for 14 of those scanners.
Localizer and axial images of ACR and body computed tomography dose index phantoms were acquired, using routine clinical techniques (120 kV and lateral localizers) on each of the 25 CT scanners: 8 GE scanners (CT750HD, VCT, and Revolution), 8 Siemens scanners (Definition AS, Force, Flash, and Edge), 5 Canon scanners (Aquilion-One, Aquilion-Prime80, and Aquilion-64), and 4 Philips scanners (iCT 256, iQon, and Ingenuity). By associating axial images with the corresponding localizer lines, the relationship between the scaled water equivalent area (WEA) and averaged LPV were established through regression analysis.
Linear relationships between the scaled WEA and the averaged LPV were observed in all 25 CT scanners ( ). Calibration parameters were similar for CT scanners from the same vendor: the coefficients of variation (COV) were ≤ 1% in all four vendor groups for the calibration slope, and < 7% for the intercept. By analyzing the deviation of WED resulted from errors in the calibration slope or intercept alone, we derived the tolerance ranges for the slope or intercept for a given WED error level. The variation of slope and intercept from different CT scanners of the same vendor introduced <±2.5% error in the estimated WED for subjects of 20 and 30-cm WED. The calibration parameters remained stable over time, with the maximum deviations all within the boundary values that introduce ±2.5% error in the estimated WED for subjects of 20 and 30-cm WED.
The stability in calibration results among CT scanners of the same vendor and over time demonstrated the feasibility of implementing WED estimation for routine clinical use.
水当量直径（WED）是一个稳健的患者体型描述符。基于定位器的 WED 估计对由于视场有限而导致的截断误差不太敏感，并且可以在轴向扫描开始之前，在一个定位器射线照片内的不同位置生成 WED 估计值。由于需要在定位器像素值（LPV）和衰减之间进行校准，并且校准结果在不同的扫描仪和随时间的稳定性未知，因此临床界认为这种方法难以实施。我们研究了来自三个医疗中心的 25 台计算机断层扫描（CT）扫描仪之间的校准结果稳定性，以及其中 14 台扫描仪在 3 至 29 个月期间的稳定性。
使用常规临床技术（120kV 和横向定位器）在 25 台 CT 扫描仪中的每一台上获取 ACR 和体部 CT 剂量指数体模的定位器和轴向图像：8 台通用电气（GE）扫描仪（CT750HD、VCT 和 Revolution）、8 台西门子（Siemens）扫描仪（Definition AS、Force、Flash 和 Edge）、5 台佳能（Canon）扫描仪（Aquilion-One、Aquilion-Prime80 和 Aquilion-64）和 4 台飞利浦（Philips）扫描仪（iCT 256、iQon 和 Ingenuity）。通过将轴向图像与相应的定位线相关联，通过回归分析建立了缩放的水等效面积（WEA）与平均 LPV 之间的关系。
在所有 25 台 CT 扫描仪中均观察到缩放的 WEA 与平均 LPV 之间存在线性关系（ ）。来自同一供应商的 CT 扫描仪的校准参数相似：对于校准斜率，四个供应商组的变异系数（COV）均≤1%，对于截距，COV<7%。通过分析仅由校准斜率或截距误差引起的 WED 偏差，我们为给定的 WED 误差水平推导了斜率或截距的公差范围。同一供应商的不同 CT 扫描仪的斜率和截距的变化会导致 20 和 30cm WED 的受试者估计 WED 时引入<±2.5%的误差。校准参数随时间保持稳定，最大偏差均在为 20 和 30cm WED 的受试者的估计 WED 引入<±2.5%的误差的边界值内。
同一供应商的 CT 扫描仪之间以及随时间的校准结果稳定性表明，实施 WED 估计用于常规临床使用是可行的。
